RyanH wrote:How much is *some* variation? My understanding is that variation should occur within the lap, but between laps I probably shouldn't do the same thing at the same points, correct? E.g. Don't hold 260w on each incline, coast and then hold 220 on the backside?
In classical "regression" testing, you do a constant speed run at, say, 15, 18, 21, 24 (and maybe 26 or 27 mph). So the range from the slowest to fastest speed is maybe 10 or more mph. In my approach, the wider the range the better because it lets you estimate both the Crr and the CdA. If you're using the same tires and tubes at the same inflation pressure and you don't really care about estimating the Crr and are just interested in the relative difference in CdA you can get away with a narrower range. The reason why a little bit of elevation change is useful is because it lets you vary the speed and power naturally. Out-and-backs on a half-pipe course are kind of handy because they let you vary the speed naturally, and they also let you vary the speed over essentially the same stretch of road since you go up and down a slope over that same section.
I'll admit that I decided to do these efforts at a constant power for partly selfish reasons, to get a hard tempo session in that I'd normally avoid, but also initially in the hopes I could see the results of my efforts by how far I covered at the end of 1 hour. Based on what you've seen, am I diluting my results because of that?
Probably fine, as long as you're not too concerned about estimating the Crr.
Lastly, a lot of people use shorter, 1km to 2km loops. Is there any issue using a 5km loop? I was assuming shorter loops are for convenience since not many want to endure a full hour to get their cda for one position.
I think the only issue is that a lot of the analytical power comes from going over the same stretch of road multiple times. If you're willing to spend an hour doing one test, there's no problem. Most of the time I don't have quite that much attention span, so I use shorter loops just so I can keep my focus. Basically, it's a mental crutch. But maybe you don't have that problem, so 5km loops are fine.
Basically, I came up with this method because I was doing lots and lots of training laps around the bike circuits at the Bois de Vincennes and the Bois de Boulogne when I lived in Paris. They were about 3.2 km around, and sometimes I'd speed up, sometimes I'd slow down, sometimes I'd do 2x20's, both courses have a little bit of elevation change so speed and power weren't constant but tended to vary, and I was wondering whether I could make use of the data to estimate my aero drag even though I couldn't do the "classical regression" method. Turns out, I could.